The Man Who Taught the World to Prompt
Most people come to frontier AI research through prestigious corridors - Stanford, MIT, DeepMind. Elvis Saravia came from Belize. A small Central American country with less than half a million people, not exactly known as a breeding ground for large language model co-creators. But then again, the AI field has always rewarded the obstinate over the credentialed.
Today, Saravia runs DAIR.AI - Democratizing Artificial Intelligence Research, Education, and Technologies - and his Prompt Engineering Guide sits at over 73,000 GitHub stars with a dedicated site at promptingguide.ai, having been used by more than three million learners. The guide did not arrive through marketing. It arrived because it was the clearest, most comprehensive resource on the subject when the subject suddenly mattered enormously.
"At this point, 'agentic engineering' has allowed me to build the best AI harness I could possibly get my hands on. You don't need to wait around for the features you need for your AI agents. Please don't. You could just build them yourself."- Elvis Saravia (@omarsar0)
Before he was teaching anyone else, he was teaching himself. A Bachelor's in Information Technology from the University of Belize, followed by four years in software development, followed by a PhD in Information Systems and Applications at National Tsing Hua University in Taiwan. Three countries, three very different worlds, one thread: the conviction that intelligent systems and the people who understand them should not be a closed club.
His PhD research focused on building empathetic AI - specifically, systems that could detect emotion and mental health signals from social media text. His 2018 paper on CARER: Contextualized Affect Representations for Emotion Recognition has since accumulated nearly 650 citations. His 2016 MIDAS paper on mental illness detection via social media added another 81. The throughline of his research is clear: AI that actually understands people, not just data.
At Meta AI, he served as a technical product marketing manager, working alongside teams including FAIR, PyTorch, and Papers with Code. He was also a co-creator of the Galactica LLM - Meta's large language model for science - which went on to receive 1,184 academic citations. It was a significant credit for anyone's resume. Saravia used it as a springboard to do his own thing.
DAIR.AI grew from a newsletter. It started as a digest called the NLP Newsletter, summarizing AI research for people who did not have time to read every paper. The format worked because Saravia writes the way researchers actually think: with rigor, but without the gatekeeping. What started as a weekly roundup has become a full education platform - courses on AI agents, prompt engineering, RAG systems, and LLMs - delivered through Maven and the DAIR.AI Academy on Thinkific. He has also consulted for leading AI companies on go-to-market strategy and product direction.
"I am at the point where my AI agents have become so good and fast at various complex tasks that I've become the bottleneck. AI agent managers are going to be in high demand soon."- Elvis Saravia (@omarsar0)
His GitHub handle is omarsar - a name that nods to his Belizean roots, distinct from the professional branding. The profile has 4,100+ followers and repositories that have collectively cleared 120,000 stars. He curates aggressively: 2,100+ starred repositories of his own, representing a living map of the AI ecosystem. He contributes in at least 10 programming languages. He speaks Spanish, English, Mandarin Chinese, and Belizean Creole.
The AI Agents Weekly newsletter - his current publishing focus - tracks the agent space with the same level of detail he once applied to NLP papers. New model releases, memory architectures, context management, long-running task patterns. The newsletter reads like dispatches from someone who is simultaneously watching the field and helping to build it. Which is exactly what he is doing.
He describes the moment of reckoning clearly: his own AI agents became so capable that he became the bottleneck. The insight is not just self-deprecating humor. It is a signal about where the field is heading - and why someone like Saravia, who has been building toward this since 2004, is positioned to help people navigate it. He is not predicting the future. He spent twenty years building toward it.